As you say, "not as first choice" - which is meaningless. It's an advertising fiction intended to create the illusion of popular support. Why stop at 50%? Just keep dropping out the lowest candidate on successive rounds until you have one candidate with 100% of the complete (meaning, ballots which ranked all candidates) votes!
As I said it's not a perfect system (and what system is?) but what it prevents is the splitting of votes between two candidates that are broadly most preferable by the majority of the voters resulting in the least popular of the three candidates being elected.
For example "Party A" and "Party B" are both roughly on the same side of the political spectrum while "Party C" is on the opposite side of the spectrum.
On election day the results are:
Party C - 32% of the vote
Party A - 30% of the vote
Party B - 28% of the vote
So in our current FPTP system the candidate for Party C wins the election with 32% of the popular vote. While the two parties on the opposite side of the political spectrum between them earned 58% of the popular vote.
Now let's hypothetically say that the vast majority (say 80%) of the last place party's voters would far prefer Party A over Party C because they are much closer in their political outlook. So under the ranked ballot system with no single party receiving over 50% of the vote Party B is dropped from the count with 80% of their supporters picking Party A as their 2nd choice and 20% picking Party C as their 2nd choice.
The results would look something like this:
Party A - 30% of vote + 22.4% 2nd choice votes from Party B (.8 x 28%) = 52.4%
Party C - 32% of vote + 5.6% 2nd choice votes from Party B (.2 x 28%) = 37.6%
Party A wins the riding with a majority of
ranked preference of 52.4% of voters.
Again, certainly not a perfect solution but better than a system where a candidate can win with the plurality of the votes but be strongly opposed by the majority of the voters in their riding.